Islet autoimmunity leading to type 1 diabetes develops

CLINICAL RESEARCH ARTICLE Impact of Age and Antibody Type on Progression From Single to Multiple Autoantibodies in Type 1 Diabetes Relatives Emanuele Bosi, 1 David C. Boulware, 2 Dorothy J. Becker, 3 Jane H. Buckner, 4 Susan Geyer, 2 Peter A. Gottlieb, 5 Courtney Henderson, 2 Amanda Kinderman, 2 Jay M. Sosenko, 6 Andrea K. Steck, 5 and Polly J. Bingley, 7 Type 1 Diabetes TrialNet Study Group 1 Diabetes Research Institute, San Raffaele Hospital and San Raffaele Vita Salute University, Milan 20132, Italy; 2 Division of Informatics and Biostatistics, University of South Florida, Tampa, Florida 33620; 3 Department of Pediatric Endocrinology, Children s Hospital of Pittsburgh, University of Pittsburgh, Pittsburgh, Pennsylvania 15224; 4 Translational Immunology Program, Benaroya Research Institute, Seattle, Washington 98101; 5 Barbara Davis Center for Childhood Diabetes, University of Colorado School of Medicine, Aurora, Colorado 80045; 6 Division of Endocrinology, University of Miami, Miami, Florida 33136; and 7 School of Clinical Sciences, University of Bristol, Bristol BS2 8DZ, United Kingdom Context: Islet autoantibodies are markers of type 1 diabetes, and an increase in number of autoantibodies detected during the preclinical phase predicts progression to overt disease. Objective: To refine the effect of age in relation to islet antibody type on progression from single to multiple autoantibodies in relatives of people with type 1 diabetes. Research Design and Methods: We examined 994 relatives with normal glucose tolerance who were positive for a single autoantibody, followed prospectively in the TrialNet Pathway to Prevention. Antibodies to glutamic acid decarboxylase (GADA), insulin (IAA), insulinoma-associated antigen 2, and zinc transporter 8 and islet cell antibodies were tested every 6 to 12 months. The primary outcome was confirmed development of multiple autoantibodies. Age was categorized as,8 years, 8 to 11 years, 12 to 17 years, and $18 years, and optimal age breakpoints were identified by recursive partitioning analysis. Results: After median follow-up of 2 years, 141 relatives had developed at least one additional autoantibodies. Five-year risk was inversely related to age, but the pattern differed by antibody type: Relatives with GADA showed a gradual decrease in risk over the four age groups, whereas relatives with IAA showed a sharp decrease above age 8 years. Recursive partitioning analysis identified age breakpoints at 14 years in relatives with GADA and at 4 years in relatives with IAA. Conclusions: In relatives with IAA, spread of islet autoimmunity is largely limited to early childhood, whereas immune responses initially directed at glutamic acid decarboxylase can mature over a longer period. These differences have important implications for monitoring these patients and for designing prevention trials. (J Clin Endocrinol Metab 102: 2881 2886, 2017) Islet autoimmunity leading to type 1 diabetes develops and progresses silently over many years before glucose intolerance and symptomatic hyperglycemia occur (1). Islet autoantibodies are the best-validated markers of this ongoing pathogenetic process and are used to predict clinical disease (2) and stage its preclinical phase (3). ISSN Print 0021-972X ISSN Online 1945-7197 Printed in USA Copyright 2017 Endocrine Society Received 5 March 2017. Accepted 11 May 2017. First Published Online 22 May 2017 Abbreviations: CI, confidence interval; GADA, antibody to glutamic acid decarboxylase; IA-2A, antibody to insulinoma-associated antigen; IAA, antibody to insulin; ICA, islet cell antibody; ZnT8A, antibody to zinc transporter 8. doi: 10.1210/jc.2017-00569 J Clin Endocrinol Metab, August 2017, 102(8):2881 2886 https://academic.oup.com/jcem 2881

2882 Bosi et al Autoantibody Progression in Pre Type 1 Diabetes J Clin Endocrinol Metab, August 2017, 102(8):2881 2886 Several prospective studies, including those following infants at genetic risk from birth, have shown that the antibody response against b cells within pancreatic islets usually targets several autoantigens, giving rise to autoantibodies to insulin (IAA), glutamic acid decarboxylase (GADA), insulinoma-associated antigen 2/ICA512 (IA2A) and zinc transporter 8 (ZnT8A), in varying sequence. Maturation of this humoral immune response, as shown by increasing autoantibody number, titer, and affinity, is associated with an increased risk for progression to the disease (4). Specifically, the number of autoantibodies detected seems crucial for the prediction of disease, with a relatively low risk associated with positivity for a single autoantibody, increasing to near certainty of development of type 1 diabetes following the appearance of multiple (i.e.,two or more) autoantibodies (4 13). Therefore, seroconversion from single to multiple autoantibodies appears to be the hallmark of a "point of no return" in the type 1 diabetes pathogenetic process, marking the transition from a state of predisposition to a preclinical stage of the disease (3). Associations have been described with younger age and HLA class II genotype (6, 10), but the determinants of this maturation remain largely unknown. Previous analyses in relatives followed prospectively in the TrialNet Pathway to Prevention (PTP) (formerly the Natural History Study) (14, 15) have indicated heterogeneity in the development and spreading of islet autoantibody responses and have also shown that progression from single to multiple autoantibodies is not restricted to early childhood and high-risk genotypes (16, 17). In light of this evidence and the implications for design of targeted interventions to prevent or delay progression of islet autoimmunity, we undertook a more in-depth investigation into the transition from single to multiple autoantibodies, with specific regard to the impact of age in relation to antibody type. Methods Study population Nondiabetic first-, second-, and third-degree relatives of people with type 1 diabetes were recruited to the TrialNet PTP (ClinicalTrials.gov identifier: NCT00097292), as previously described (14). All study participants gave informed consent, and the Ethics Committee responsible for each clinical site approved the study. Participants were included in this analysis if they had antibodies to the same single islet autoantigen (GADA, IAA, or IA-2A) detected on at least two occasions, and antibody results were available from at least one subsequent study visit. All samples were screened for GADA, IAA, and IA-2A. If levels of any of these were above the threshold of positivity, samples were additionally tested for islet cell antibodies (ICAs) and ZnT8A. Individuals with confirmed islet autoantibodies underwent baseline assessment, including oral glucose tolerance testing, and were followed every 6 to 12 months in accordance with the PTP study protocol, as previously reported (14). Relatives with the protective HLA DQB1*0602 allele were not included in this analysis. Assays GADA, IAA, IA-2A, and ZnT8A were measured by radioimmunoassay in the TrialNet Core laboratory at the Barbara Davis Center for Childhood Diabetes, Denver, Colorado, and ICA was measured by indirect immunofluorescence at the University of Florida, Gainesville, Florida, as previously described (18 20). Statistical analysis The primary outcome of the analysis was confirmed development of multiple autoantibodies, defined as detection on two occasions of at least two of the five islet autoantibodies included in the testing strategy (GADA, IAA, IA-2A, ZnT8A, and ICA); this confirmation is based on two consecutive autoantibody tests that are done within 1 year of each other. The time-to-event was calculated from the date of first detection of a single islet autoantibody to date of first detection of multiple autoantibodies. The risk for developing multiple autoantibodies was assessed by survival analysis using Kaplan-Meier curves. Cox proportional hazards regression models were used for multivariable analysis. Age groups were initially categorized by using boundaries based on common definitions of infancy to prepuberty, adolescence, and adulthood (,8 years, 8 to 11 years, 12 to 17 years, and $18 years) and then refined by recursive partitioning analysis used to identify optimal age breakpoints within each single autoantibody population (21 23). Recursive partitioning is a model-based method used to identify a cutpoint for a marker that best differentiates subjects in relation to an outcome of interest, such as risk of progression to multiple positive autoantibodies. All P values were two-sided and statistical significance was determined using a threshold of 0.05. The statistical program SAS (version 9.2 for Windows; SAS Institute, Cary, NC) was used for all primary analyses including assessment of baseline characteristics and time to event analyses. In addition, the statistical program R (version 3.1.2 for Windows; R Foundation for Statistical Computing, Vienna, Austria) was used in analyses for identifying optimal cut-points, specifically, we utilized recursive partitioning analyses. Results Of 151,458 relatives screened in the TrialNet PTP between 1 March 2004 and 31 March 2015, 994 were positive for a single autoantibody (GADA, IAA, or IA- 2A) with normal oral glucose tolerance at baseline and were therefore eligible for inclusion in the analysis; an additional 276 persons had abnormal glucose tolerance. Of the 994 positive for a single autoantibody with normal glucose tolerance, 709 (71.3%) had GADA, 236 (23.7%) had IAA, and 49 (4.9%) had IA-2A; 59.6% were female. The median age of the participants was 17.6 years (interquartile range, 9.8 to 36.2 years); 183 (18.4%) were younger than 8 years of age, 157 (15.8%) were age 8 to 11 years, 169 (17.0%) were 12 to 17 years old, and 485 (48.8%) were age 18 years or older.

doi: 10.1210/jc.2017-00569 https://academic.oup.com/jcem 2883 After a median follow-up of 2.0 years (interquartile range, 0.8 to 3.8 years), 141 relatives had developed at least one additional autoantibody. Estimated cumulative risk within 5 years was 23% [95% confidence interval (CI), 19% to 27%] overall and did not vary among autoantibody types [GADA, 25% (95% CI, 20% to 30%); IAA, 19% (95% CI, 11% to 27%); and IA-2A, 23% (95% CI, 7% to 38%); P = 0.09]. The overall risk for developing additional autoantibodies was inversely related to age (multivariable hazard ratio, 0.96; 95% CI, 0.94 to 0.99; P = 0.005), but heterogeneity among autoantibody type was identified. Table 1 shows the risk for developing multiple autoantibodies within 5 years, categorized by age. Relatives with GADA showed a gradual decrease in risk across the four age groups, whereas relatives with IAA showed a sharp decrease in risk above age 8 years. Analysis between age groups was not performed in relatives with IA-2A because of an insufficient number of participants. The age distribution of the relatives who developed multiple autoantibodies also differed between those with GADA and those with IAA (Fig. 1). Relatives younger than 8 years of age represented 71% of the IAA-positive individuals who became multiple autoantibody positive within 5 years, whereas the ages of GADA-positive individuals who progressed were more evenly distributed throughout childhood and adolescence. Of the relatives who developed one or more additional autoantibodies, 17 of 24 (71%) with IAA vs 22 of 99 (22%) with GADA were younger than age 8 years (P = 0.048). Recursive partitioning analysis confirmed differences in age-related risk between GADA-positive and IAApositive groups and identified age breakpoints at 14 years in relatives with GADA alone and 4 years in those with IAA alone. The survival time from initial detection of a single islet autoantibody to first detection of multiple antibodies in subgroups categorized by these age breakpoints is shown in Fig. 2(a) for GADA-positive relatives and Fig. 2(b) for IAA-positive relatives. Among the GADA-positive relatives, 99 individuals developed additional autoantibodies within 5 years, of whom 59 were aged younger than 14 years and 40 were aged 14 years or older [estimated cumulative risk, 35.2% (95% CI, 27.9% to 43.8%) vs 17.7% (95% CI, 12.6% to 24.6%), respectively; log-rank test P, 0.001]. Among IAA-positive relatives, 24 individuals developed additional autoantibodies within 5 years, of whom 12 were younger than age 4 years and 12 were 4 years of age or older [estimated cumulative risk, 73.1% (95% CI, 48.5% to 92.6%) vs 11.4% (95% CI, 6.3% to 20.4%), respectively; log-rank test, P, 0.001]. Discussion The main finding of this analysis is that the interaction between age and risk for progression from single to multiple autoantibodies differs according to whether the primary autoantigen is glutamic acid decarboxylase or insulin. Specific age breakpoints were very different in those initially positive for IAA compared with those with GADA. This suggests that early interventions targeting single autoantibody positive individuals at risk for type 1 diabetes may differ in their effectiveness at different ages, depending on whether they have GADA or IAA, and provides some guidance as to what age groups to target. Previous analyses in relatives followed prospectively in the TrialNet PTP (14, 15) found that in those positive for a single autoantibody, the risk for progression to multiple antibodies was 22% within 5 years and to type 1 diabetes was 6% (16). As confirmed in the present analysis, overall risk for development of multiple autoantibodies was independent of antibody type, inversely related to age, and associated with high- and intermediate-risk HLA class II genotypes and high GADA titers (16). In addition, progression appeared not to be influenced by initial body mass index or other metabolic variables (24). More recently, recursive partitioning analysis demonstrated that age and GADA titer taken together were helpful in stratifying the overall risk for progression from single to multiple autoantibodies (17). Table 1. Risk for Developing Multiple Autoantibodies Within 5 Years in Single Autoantibody Positive Relatives Categorized by Age Age at Entry, y Antibody Type <8 8 11 12 17 >18 Overall GADA, n 107 113 117 372 709 Risk, % 35 (24 46) 38 (24 51) 28 (14 41) 16 (10 22) 25 (20 30) IAA, n 69 33 43 91 236 Risk, % 37 (22 52) 4 (0 11) 13 (0 34) 13 (0 25) 19 (11 27) IA-2A, n 7 11 9 22 49 Risk, % Insufficient number of participants to assess 23 (7 38) Values in parentheses are 95% CIs.

2884 Bosi et al Autoantibody Progression in Pre Type 1 Diabetes J Clin Endocrinol Metab, August 2017, 102(8):2881 2886 Figure 1. Particiants who develped additional autoantibodies within 5 years, subdivided by age group. (a) GADA-positive (n = 99) participants. (b) IAA-positive participants (n = 24). Among the relatives who progressed, 17 of 24 (71%) with IAA vs 27 of 99 (27%) with GADA were younger than 8 years of age (P, 0.05). These additional analyses focus on evaluation of the time course of progression among single autoantibody positive persons, depending on whether the first autoantibodies detected were GADA or IAA. These demonstrate marked differences between the initial IAA and GADA groups. Although risk was inversely related to age in both groups, the relationships were not the same; whereas risk in GADA-positive relatives decreased gradually up to age 18 years, in IAA-positive relatives the risk was concentrated in younger children, with high risk for progression up to age 8 years and a sharp decline thereafter. This pattern was confirmed by recursive partitioning analysis, which showed significant differences in risk and identified different age breakpoints for IAA (4 years) and for GADA (14 years). On the basis of these observations, spread of autoimmunity from insulin to other islet antigens seems largely limited to early childhood, whereas autoimmune responses initially directed against glutamic acid decarboxylase can mature over a longer period, including the whole of adolescence. It will be interesting to see whether this pattern is confirmed in studies of cellular immune responses. These findings are relevant to the design of prevention trials targeting the early phases of type 1 Figure 2. Time from initial detection of (a) single GADA to first detection of multiple antibodies in relatives,14 years (black line) and $14 years (gray line) and (b) single IAA to first detection of multiple antibodies in relatives,4 years (black line) and $4 years (gray line). Aab, autoantibody; Ab, antibody. diabetes associated autoimmunity. In individuals with IAA alone, intervention should be considered only below age 8 years, with the highest priority given to individuals younger than 4 years of age; conversely, in individuals with GADA alone, intervention seems justified up to age 18 years, with the highest potential below 14 years. Although there is no sharp decline after age 18, the overall risk for developing additional antibodies is relatively low in GADA-positive adults (16% within 5 years), and we suggest that issues of power and the sample size required should be carefully evaluated before these persons are included in trials of type 1 diabetes prevention. A particular strength of this study is the size of the cohort, which was screened and followed up every 6 to 12 months according to a standard protocol (14). One limitation is that, in contrast with studies that followed infants from birth, the time at which seroconversion has occurred is unknown when relatives are found to be positive for an autoantibody in the TrialNet PTP. This may lead to underestimation of the duration of single autoantibody positivity; however, because this study

doi: 10.1210/jc.2017-00569 https://academic.oup.com/jcem 2885 focused on time to further seroconversion to multiple autoantibodies, the time of initial seroconversion is less relevant in this context. As in other TrialNet studies, we have included appearance of ICA in the definition of development of multiple autoantibodies because although GADA and IA-2A contribute to ICA staining (25, 26), overlap is incomplete and previous analyses in the PTP cohort have demonstrated that ICA is associated with additional risk (20). Although the determinants of islet autoimmunity underlying type 1 diabetes remain unknown, accumulating evidence suggests that it develops through multistep involvement of several pathogenetic pathways (27). Accordingly, interventions at an earlier stage might rely on simpler approaches, such as antigen-based therapies, whereas at a later stage, a more complex approach (based, for instance, on combination therapies) is likely to be necessary to halt or delay the progression of the disease process (28). On the basis of our current ability to predict type 1 diabetes, single autoantibody positivity is the earliest detectable sign of the ongoing autoimmune process, when the chances of a successful intervention could be greatest. The evidence of heterogeneity in the progression of islet autoimmunity in relatives of the TrialNet PTP study, the largest study cohort ever screened and followed up, calls for different therapeutic approaches in single autoantibody positive individuals with IAA or GADA, with the highest potential for success in interventions designed for early childhood in IAA-positive and no later than adolescence in GADA-positive individuals. Acknowledgments Address all correspondence and requests for reprints to: Emanuele Bosi, MD, Diabetes Research Institute, San Raffaele Hospital and Vita-Salute University, Via Olgettina 60, 20132 Milan, Italy. E-mail: bosi.emanuele@hsr.it. The sponsor of the trial was the Type 1 Diabetes TrialNet Study Group. This group is a clinical trials network funded by the National Institutes of Health (NIH) through the National Institute of Diabetes and Digestive and Kidney Diseases, the National Institute of Allergy and Infectious Diseases, and the Eunice Kennedy Shriver National Institute of Child Health and Human Development, through Cooperative Agreements U01 DK061010, U01 DK061034, U01 DK061042, U01 DK061058, U01 DK085465, U01 DK085453, U01 DK085461, U01 DK085466, U01 DK085499, U01 DK085504, U01 DK085509, U01 DK103180, U01 DK103153, U01 DK085476, U01 DK103266, U01 DK103282, U01 DK106984, U01 DK106994, U01 DK107013, U01 DK107014, and UC4 DK106993, and the Juvenile Diabetes Research Foundation International (JDRF). The contents of this article are solely the responsibility of the authors and do not necessarily represent the official views of the NIH or the JDRF. Author contributions: All authors were members of the TrialNet Study Group and contributed to the data used in this article. E.B. and P.J.B. wrote the manuscript. P.J.B. and D.C.B. designed and conducted the statistical analysis. All authors contributed to discussion, reviewed/edited the manuscript, and gave final approval for the paper to be published. S.G. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Clinical trial registry: ClinicalTrials.gov no. NCT00097292 (registered 19 November 2004). A complete list of the Type 1 Diabetes TrialNet Study Group can be found in the Supplementary Data. Disclosure Summary: The authors have nothing to disclose. References 1. Atkinson MA, Eisenbarth GS, Michels AW. Type 1 diabetes. Lancet. 2014;383(9911):69 82. 2. Bonifacio E. Predicting type 1 diabetes using biomarkers. Diabetes Care. 2015;38(6):989 996. 3. Insel RA, Dunne JL, Atkinson MA, Chiang JL, Dabelea D, Gottlieb PA, Greenbaum CJ, Herold KC, Krischer JP, Lernmark Å, Ratner RE, Rewers MJ, Schatz DA, Skyler JS, Sosenko JM, Ziegler AG. Staging presymptomatic type 1 diabetes: a scientific statement of JDRF, the Endocrine Society, and the American Diabetes Association. Diabetes Care. 2015;38(10):1964 1974. 4. Ziegler AG, Rewers M, Simell O, Simell T, Lempainen J, Steck A, Winkler C, Ilonen J, Veijola R, Knip M, Bonifacio E, Eisenbarth GS. Seroconversion to multiple islet autoantibodies and risk of progression to diabetes in children. JAMA. 2013;309(23):2473 2479. 5. Orban T, Sosenko JM, Cuthbertson D, Krischer JP, Skyler JS, Jackson R, Yu L, Palmer JP, Schatz D, Eisenbarth G; Diabetes Prevention Trial- Type 1 Study Group. Pancreatic islet autoantibodies as predictors of type 1 diabetes in the Diabetes Prevention Trial-Type 1. Diabetes Care. 2009;32(12):2269 2274. 6. Ziegler AG, Bonifacio E; BABYDIAB-BABYDIET Study Group. Age-related islet autoantibody incidence in offspring of patients with type 1 diabetes. Diabetologia. 2012;55(7):1937 1943. 7. SteckAK,JohnsonK,BarrigaKJ,MiaoD,YuL,HuttonJC,Eisenbarth GS, Rewers MJ. Age of islet autoantibody appearance and mean levels of insulin, but not GAD or IA-2 autoantibodies, predict age of diagnosis of type 1 diabetes: diabetes autoimmunity study in the young. Diabetes Care. 2011;34(6):1397 1399. 8. Parikka V, Näntö-Salonen K, Saarinen M, Simell T, Ilonen J, Hyöty H, Veijola R, Knip M, Simell O. Early seroconversion and rapidly increasing autoantibody concentrations predict prepubertal manifestation oftype1diabetesinchildrenatgeneticrisk. Diabetologia. 2012;55(7): 1926 1936. 9. Steck AK, Vehik K, Bonifacio E, Lernmark A, Ziegler AG, Hagopian WA, She J, Simell O, Akolkar B, Krischer J, Schatz D, Rewers MJ; TEDDY Study Group. Predictors of Progression From the Appearance of Islet Autoantibodies to Early Childhood Diabetes: The Environmental Determinants of Diabetes in the Young (TEDDY). Diabetes Care. 2015;38(5):808 813. 10. Krischer JP, Lynch KF, Schatz DA, Ilonen J, Lernmark Å, Hagopian WA, Rewers MJ, She JX, Simell OG, Toppari J, Ziegler AG, Akolkar B, Bonifacio E; TEDDY Study Group. The 6 year incidence of diabetes-associated autoantibodies in genetically atrisk children: the TEDDY study. Diabetologia. 2015;58(5): 980 987. 11. Achenbach P, Hummel M, Thümer L, Boerschmann H, Höfelmann D, Ziegler AG. Characteristics of rapid vs slow progression to type 1 diabetes in multiple islet autoantibody-positive children. Diabetologia. 2013;56(7):1615 1622.

2886 Bosi et al Autoantibody Progression in Pre Type 1 Diabetes J Clin Endocrinol Metab, August 2017, 102(8):2881 2886 12. Chmiel R, Giannopoulou EZ, Winkler C, Achenbach P, Ziegler AG, Bonifacio E. Progression from single to multiple islet autoantibodies often occurs soon after seroconversion: implications for early screening. Diabetologia. 2015;58(2):411 413. 13. Steck AK, Dong F, Waugh K, Frohnert BI, Yu L, Norris JM, Rewers MJ. Predictors of slow progression to diabetes in children with multiple islet autoantibodies. JAutoimmun. 2016;72: 113 117. 14. Mahon JL, Sosenko JM, Rafkin-Mervis L, Krause-Steinrauf H, Lachin JM, Thompson C, Bingley PJ, Bonifacio E, Palmer JP, Eisenbarth GS, Wolfsdorf J, Skyler JS; TrialNet Natural History Committee; Type 1 Diabetes TrialNet Study Group. The TrialNet Natural History Study of the Development of Type 1 Diabetes: objectives, design, and initial results. Pediatr Diabetes. 2009; 10(2):97 104. 15. Vehik K, Beam CA, Mahon JL, Schatz DA, Haller MJ, Sosenko JM, Skyler JS, Krischer JP; TrialNet Natural History Study Group. Development of autoantibodies in the TrialNet Natural History Study. Diabetes Care. 2011;34(9):1897 1901. 16. Bingley PJ, Boulware DC, Krischer JP; Type 1 Diabetes TrialNet Study Group. The implications of autoantibodies to a single islet antigen in relatives with normal glucose tolerance: development of other autoantibodies and progression to type 1 diabetes. Diabetologia. 2016;59(3):542 549. 17. Xu P, Krischer JP; Type 1 Diabetes TrialNet Study Group. Prognostic classification factors associated with development of multiple autoantibodies, dysglycemia, and type 1 diabetes-a recursive partitioning analysis. Diabetes Care. 2016;39(6):1036 1044. 18. Yu L, Rewers M, Gianani R, Kawasaki E, Zhang Y, Verge C, Chase P, Klingensmith G, Erlich H, Norris J, Eisenbarth GS. Antiislet autoantibodies usually develop sequentially rather than simultaneously. J Clin Endocrinol Metab. 1996;81(12):4264 4267. 19. Bonifacio E, Yu L, Williams AK, Eisenbarth GS, Bingley PJ, Marcovina SM, Adler K, Ziegler AG, Mueller PW, Schatz DA, Krischer JP, Steffes MW, Akolkar B. Harmonization of glutamic acid decarboxylase and islet antigen-2 autoantibody assays for national institute of diabetes and digestive and kidney diseases consortia. J Clin Endocrinol Metab. 2010;95(7):3360 3367. 20. Yu L, Boulware DC, Beam CA, Hutton JC, Wenzlau JM, Greenbaum CJ, Bingley PJ, Krischer JP, Sosenko JM, Skyler JS, Eisenbarth GS, Mahon JL; Type 1 Diabetes TrialNet Study Group. Zinc transporter-8 autoantibodies improve prediction of type 1 diabetes in relatives positive for the standard biochemical autoantibodies. Diabetes Care. 2012;35(6):1213 1218. 21. Ciampi IA, Thiffault J. Recursive partition and amalgamation (RECPAM) for censored survival data: criteria for tree selection. Statistical Software Newsletter. 1988;14:78 81. 22. Gordon L, Olshen RA. Tree-structured survival analysis. Cancer Treat Rep. 1985;69(10):1065 1069. 23. Breiman L, Friedman JH, Olshen RA, Stone CJ. Classification and Regression Trees. Belmont, CA: Wadsworth International Group; 1984. 24. Meah FA, DiMeglio LA, Greenbaum CJ, Blum JS, Sosenko JM, Pugliese A, Geyer S, Xu P, Evans-Molina C; Type 1 Diabetes TrialNet Study Group. The relationship between BMI and insulin resistance and progression from single to multiple autoantibody positivity and type 1 diabetes among TrialNet Pathway to Prevention participants. Diabetologia. 2016;59(6):1186 1195. 25. Richter W, Seissler J, Northemann W, Wolfahrt S, Meinck HM, Scherbaum WA. Cytoplasmic islet cell antibodies recognize distinct islet antigens in IDDM but not in stiff man syndrome. Diabetes. 1993;42(11):1642 1648. 26. Bonifacio E, Lampasona V, Genovese S, Ferrari M, Bosi E. Identification of protein tyrosine phosphatase-like IA2 (islet cell antigen 512) as the insulin-dependent diabetes-related 37/40K autoantigen and a target of islet-cell antibodies. J Immunol. 1995;155(11): 5419 5426. 27. Bluestone JA, Herold K, Eisenbarth G. Genetics, pathogenesis and clinical interventions in type 1 diabetes. Nature. 2010;464(7293): 1293 1300. 28. Skyler JS. Prevention and reversal of type 1 diabetes past challenges and future opportunities. Diabetes Care. 2015;38(6):997 1007.